This document discusses treatment strategies for cardioesophageal cancer (CEC). It analyzes data from 175 patients who underwent radical esophagogastrectomy for CEC between 1975-2008. It finds that 5-year survival was significantly better for patients who received adjuvant chemotherapy and immunotherapy after surgery compared to surgery alone. The best treatment strategies for CEC are identified as: 1) early detection of CEC, 2) surgery by experienced surgeons given the complexity, 3) aggressive surgery with adequate lymph node dissection, 4) adjuvant chemotherapy with immunotherapy to improve prognosis, and 5) high precision in predicting patient outcomes.

1. The Open Cardiovascular and Thoracic Surgery Journal, 2009, 2, 21-32 21
Open Access
Cardioesophageal Cancer: Best Treatment Strategies
Oleg Kshivets*
Department of Thoracic Surgery, Klaipeda University Hospital, Klaipeda, Lithuania
Abstract: Objective: We examined the clinicomorphologic factors associated with the low- and high-risk of generaliza-
tion of cardioesophageal cancer (CEC) (T1-4N0-2M0) after complete (R0) esophagogastrectomies (EG) through left tho-
racoabdominal incision.
Methods: We analyzed data of 175 consecutive CEC patients (CECP) (age = 55.3 ± 8.7 years; tumor size = 6.9 ± 3.3 cm)
radically operated and monitored in 1975-2008 (males = 132, females = 43; combined EG with resection of pancreas,
liver, diaphragm, colon transversum, lung, trachea, pericardium, splenectomy - 71; lymphadenectomy D2 - 98, D3-D4 -
77; esophagogastroanastomosis - 98, esophagoenteroanastomosis - 77; adenocarcinoma - 112, squamous cell carcinoma -
58, mix - 5; T1 - 24, T2 - 38, T3 - 66, T4 - 47; N0 - 70, N1 - 22, N2 - 83; G1 - 52, G2 - 34, G3 - 89; surgery alone - 128;
surgery and adjuvant chemoimmunotherapy-AT: 5FU + thymalin/taktivin, 5-6 cycles - 47 CECP).
Variables selected for 5-year survival (5YS) study were input levels of 45 blood parameters, sex, age, TNMPG, cell type,
tumor size. Survival curves were estimated by the Kaplan-Meier method. Differences in curves between groups of CECP
were evaluated using a log-rank test. Multivariate Cox modeling, multi-factor clustering, structural equation modeling,
Monte Carlo, bootstrap simulation and neural networks computing were used to determine any significant dependence.
Results: For total of 175 CECP overall life span (LS) was 1381.6 ± 1486.7 days, (median - 723 days) and cumulative 5YS
reached 36.1%, 10 years - 26.6%. 53 CECP lived more than 5 years without CEC progressing. 104 CECP died because of
CEC during the first 5 years after surgery. 5YS of CECP was superior significantly after AT (69.9%) compared with sur-
gery alone (26.6%) (P = 0.000 by log-rank test). Cox modeling displayed that 5YS of CECP (n = 175) after complete EG
significantly depended on: phase transition (PT) early-invasive CEC, PT N0-N12, AT, age, T, tumor growth, Rh-factor,
blood cell subpopulations, cell ratio factors (P = 0.000-0.047). Neural networks computing, genetic algorithm selection
and bootstrap simulation revealed relationships between 5YS and early-invasive CEC (rank = 1), PT N0-N12 (rank = 2),
AT (rank = 3), T (4), gender (5), prothrombin index (6), weight (7), glucose (8), age (9), coagulation time (10), eosino-
phils/cancer cells (11), erythrocytes/cancer cells (12), hemorrhage time (13), protein (14), Hb (15), segmented neutro-
phils/cancer cells (16), stab neutrophils/cancer cells (17). Correct prediction of 5YS was 100% by neural networks com-
puting (error = 0.0009e-12; urea under ROC curve = 1.0).
Conclusions: Best treatment strategies for CECP are: 1) screening and early detection of CEC; 2) availability of very ex-
perienced thoracoabdominal surgeons because of complexity of radical procedures; 3) aggressive en block surgery and
adequate abdominal, mediastinal, cervical lymphadenectomy for completeness; 4) high-precision prediction; 5) adjuvant
chemoimmunotherapy for CECP with unfavorable prognosis.
INTRODUCTION gastric cancer [4]. Formal definition of cardia, given by J.
Siewert, as on 1 cm is higher and 2 cm below gastro-
The problem of cardioesophageal cancer (CEC) took on esophageal junction, served by the basis for his classification
special significance last years. This is caused by 300-500%
of CEC as the tumors located on 5 cm is higher and below
growth of this fatal disease during last 30 years, having char-
anatomic cardia, in any way does not reflect the features of
acter of a pandemic in the USA, European Union, Russia,
this disease [5,6]. And they are rather essential. The progno-
Japan, etc. [1,2]. Today C C takes the 6th place in cancer
sis of CEC is much worse significantly, than esophageal
mortality in the world. It ranks special position in medicine
cancer or gastric cancer alone, that is caused by extreme ag-
and represents a fundamental challenge for surgery. The real gressive behavior and very fast local and system dissemina-
5-year survival (5YS) across all stages of CEC is approxi-
tion of CEC since this oncopathology is located on the bor-
mately 5-10% in the USA and Europe [2, 3]. Disputes begin
der of two cavities (abdomen and thorax) and two mucous
already at the definition of CEC concept. Actually, valid
coats (esophagus and stomach). Agreeably, lymph outflow
classification of CEC is absent, as a result of a cardia cancer
occurs in two regions (abdominal and mediastinal) (Fig. 1).
with primary spreading on the esophagus classifies as an
Infiltration of cancer cells upwards on the gullet can differ
esophageal cancer, and with expansion to the stomach - as a on 10-12 cm from visible tumor burdens, achieving some-
times till cervical part of esophagus [7]. In view of close
*Address correspondence to this author at the Thoracic Surgery Department,
mutual relation of cardioesophageal junction to several or-
Klaipeda University Hospital, Vingio: 16, P/D 1017, Klaipeda, LT95188, gans (stomach, gullet, liver, pancreas, lien, diaphragm, lung,
Lithuania; Tel: (370)60878390; E-mail: kshivets003@yahoo.com pericardium, colon transversum) completeness of procedures
for local advanced CEC can be achieved only by multiorgans
1876-5335/09 2009 Bentham Open

2. 22 The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 Oleg Kshivets
resections. Therefore treatment of tumors for this localiza-
tion is one of the most difficult and disputable sections of
thoracoabdominal surgery [3]. Extensive abdominal (D2-
D4), mediastinal, and, in some cases, cervical lymphadenec-
tomy (2-3 folds) is the cornerstone of the radical
esophagogastrectomy [8]. It has not only medical and diag-
nostic, but also prognostic value. All this demands from the
surgeon of virtuosous techniques, huge experience, force and
endurance. That is why the surgery of CEC always remains
the privilege of a narrow circle of the best surgeons of the
world, which deficiency accrues every year. And finally,
usually the basic attention during analysis of CEC was given
to the nearest postoperative results in view of complexity of
surgical interventions and features of postoperative monitor-
ing of patients, and the remote period remains in a shade.
Considering the premises, we analyzed the long time results
of treatment of patients with CEC (CECP) after radical (R0)
esophagogastrectomies through the left thoracoabdominal
incision (Garlock operation). We developed best treatment
strategies that incorporate bolus chemotherapy and immuno-
therapy after radical, aggressive en-block surgery and ade-
quate abdominal, mediastinal and cervical lymph node dis-
section.
PATIENTS AND METHODS
We performed a review of prospectively collected data-
base of European patients undergoing the complete (R0)
esophagogastrectomy for CEC between 1975 and 2008. 175
consecutive CECP (male - 132, female - 43; age = 55.3 ± 8.7
years, tumor size = 6.9 ± 3.3 cm) (mean ± standard devia-
tion) entered this trial. Patients were not considered eligible
if they had stage IV (nonregional lymph nodes metastases
and distant metastases), previous treatment with chemother-
apy, immunotherapy or radiotherapy or if there were two
primary tumors at the time of diagnosis. CECP after non-
radical procedures and patients, who died postoperatively, Fig. (1). Scheme of lymph node metastasis for cardioesophageal
were excluded to provide a homogeneous patient group. The cancer: 1 - right paracardial; 2 - left paracardial; 7 - along left gas-
preoperative staging protocol included clinical history, tric artery; 8 - along common hepatic artery; 9 - along celiac trunk;
physical examination, complete blood count with differen- 10 - lien hilar; 11 - along lienal artery; 16 - aortocaval and paraaor-
tials, biochemistry and electrolyte panel, chest X-rays, rönt- tal abdominal; 104a - right supraclavicular; 104b - left supraclavi-
genoesophagogastroscopy, computed tomography scan of cular; 105 - upper paraesophageal; R106 - right paratracheal; L106
thorax, abdominal ultrasound, fibroesophagogastroscopy, - left paratracheal; 107 - bifurcational; 108 - middle paraesophag-
electrocardiogram. Computed tomography scan of abdomen, eal; 109 - lung hilar; 110 - lower paraesophageal; 111 - diaphrag-
liver and bone radionuclide scan were performed whenever matic; 112 - paraaortal thoracic.
needed. Mediastinoscopy was not used. All CECP were di-
with preservation of right gastroepiploic vessels or total
agnosed with histologically confirmed CEC. All had meas- stomach with an esophagus on 6-12 m above proximal tu-
urable tumor and ECOG performance status 0 or 1. Before mor burden with intraoperative express-histology for clear-
any treatment each patient was carefully examined by a ance without fail, minor and major omentum, and also lymph
medical panel composed of thoracic surgeon, chemothera-
nodes on the both sides of a diaphragm (lymph nodes along
peutist, radiologist and gastroenterologist to confirm the
of celiac trunk, common hepatic artery, splenic artery toward
stage of disease. All patients signed a written informed con-
the splenic hilum, paracardial, paraesophageal, sub- and su-
sent form approved by the local Institutional Review Board. pradiaphragmatic, mediastinal, bifurcational, paraaortal and
The initial treatment was started with surgery. We are aortocaval lymph nodes) (Figs. 1, 3). The left gastric artery
basic supporters of Garlock’s operation through left thoraco- was always transected at its origin and remained with the
abdominal incision for CEC as more convenient and less specimen. The present analysis was restricted to CECP with
traumatic, than Lewis's operation (Figs. 2, 3). Conclusive complete resected tumors with negative surgical resection
advantages of left thoracoabdominal approach are unsur- margin and with N0-2 lymph nodes. Patients with the CEC
passed conveniences for combined multiorgans resections, infiltration till upper third of esophagus had an additional
extensive lymphadenectomy and exact correlation of length formal extended mediastinal lymphadenectomy comprising
of a gastric tube or jejunum with the level of anastomosis [9- all nodes at the tracheal bifurcation along the left and right
12]. At this procedure we removed the most part of stomach main stem bronchi, aorta window, the upper mediastinal
compartment, and along the left recurrent nerve. A system-

3. Cardioesophageal Cancer: Best Treatment Strategies The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 23
Fig. (2). Scheme of left thoracoabdominal incision and Garlock procedure.
atic cervical lymphadenectomy was performed routinely for chemoimmunotherapy was accomplished in CECP with
CECP with neck anastomosis. 98 patients underwent lymph ECOG performance status 0 or 1.
nodal D2-dissection (in terms of gastric cancer surgery).
All patients (175 CECP) were divided randomly between
Extensive lymph nodal D3-D4-dissection was performed in
the two protocol treatment: 1) surgery and adjuvant chemo-
77 CECP. Routine two-field lymphadenectomy (in terms of immunotherapy (47 CECP - group A); 2) surgery alone
esophageal surgery) was performed in 133, three-field - in
without any adjuvant treatment (128 CECP - group B) - the
42. Complete surgical procedure consisted of
control group.
esophagogastrectomy with one-stage intrapleural
esophagogastrostomy or esophagoenterostomy in 133 and Forty-seven CECP received adjuvant chemoimmunother-
with anastomosis on the neck - in 42. CEC was localized till apy which consisted of chemoimmunotherapy (5-6 cycles)
lower third of esophagus in 85, middle third - in 61, upper (group A). 1 cycle of bolus chemotherapy was initiated 3-5
third - in 29. Among these, 71 CECP underwent combined weeks after complete esophagogastrectomies and consisted
esophagogastrectomy with multiorgans resections of pan- of fluorouracil 500 mg/m2 intravenously for 5 days. Immu-
creas, liver, diaphragm, colon transversum, lung, trachea, notherapy consisted thymalin or taktivin 20 mg intramuscu-
pericardium, splenectomy. Esophagogastric and esophagoje- larly on days 1, 2, 3, 4 and 5. These immunomodulators pro-
junic anastomosis were carried out manually as an inktype duced by Pharmaceutics of Russian Federation (Novosibirsk)
“end to end”. For total gastrectomy we added lymph node and approved by Ministry of Health of Russian Federation.
dissection of subpyloric, lienal, retropancreatic, hepato- Thymalin and taktivin are preparations from calf thymus,
duodenal, aortocaval, supramesenteric, mesocolonic lymph which stimulate proliferation of blood T-cell and B-cell sub-
nodes (Fig. 1). Complete resection (R0) was defined as re- populations and their response [13]. The importance must be
moval of the primary tumor and all accessible tissues, lymph stressed of using immunotherapy in combination with che-
nodes, with no residual tumor left behind (resection of all motherapy, because immune dysfunctions of the cell-
macroscopic tumor and resection margins free of tumor at mediated and humoral response were induced by tumor, sur-
microscopic analysis). gical trauma and chemotherapy [14]. Such immune defi-
ciency induced generalization of CEC and compromised the
All CECP were postoperatively staged according to the
long-term surgery result. In this sense, immunotherapy may
TNMG-classification. Histological examination showed
have shielded the patient from adverse side effects of treat-
adenocarcinoma in 112, squamous cell carcinoma - in 58 and
ment. During chemoimmunotherapy antiemetics were ad-
mixed carcinoma - in 5 patients. The pathological T stage ministered. Gastrointestinal side effects, particularly nausea
was T1 in 24, T2 - in 38, T3 - in 66, T4 - in 47 cases; the
and vomiting, were mild, and chemoimmunotherapy was
pathological N stage was N0 in 70, N1 - in 22, N2 - in 83
generally well tolerated. Severe leukopenia, neutropenia,
patients. The tumor differentiation was graded as G1 in 52,
anemia and trombocytopenia occurred infrequently. There
G2 - in 34, G3 - in 89 cases. After surgery postoperative
were no treatment-related deaths.

4. 24 The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 Oleg Kshivets
Fig. (3). Our methodology of esophagogastrectomy through left thoracoabdominal approach.
A follow-up examination was, generally, done every 3 Inc., Needham, MA, USA), SIMSTAT (Provalis Research,
month for the first 2 years, every 6 month after that and Inc., Montreal, QC, Canada). All tests were considered sig-
yearly after 5 years, including a physical examination, a nificant if the resulting P value was less than 0.05.
complete blood count, blood chemistry, and chest roent-
genography. Fibroesophagogastroscopy was done every 6 RESULTS
months for the first 3 years. Zero time was the data of surgi- For the entire sample of 175 patients overall LS was
cal procedures. No one was lost during the follow-up period 1381.6 ± 1486.7 days (mean ± standard deviation) (95% CI,
and we regarded the outcome as death through personal 1159.8-1603.4 days; median = 723). General cumulative
knowledge, physician's reports, autopsy or death certificates. 5YS reached 36.1%, 10-year survival - 26.6%. 61 CECP
Survival time (days) was measured from the date of surgery (34.9%) were alive till now, 53 CECP (30.3%) lived more
until death or the most-recent date of follow-up for surviving than 5 years (LS = 3212.7 ± 1512.0 days) without any fea-
patients. tures of CEC progressing. 104 CECP (59.4%) died because
Variables selected for 5YS and life span (LS) study were of generalization during the first 5 years after surgery (LS =
the input levels of 45 blood parameters, sex, age, TNMG, 587.3 ± 316.8 days) (Fig. 4).
cell type, and tumor size. Survival curves were estimated by It is necessary to pay attention to the two very important
the Kaplan-Meier method. Differences in curves between prognostic phenomenas. First, we found 100% 5YS for
groups of CECP were evaluated using a log-rank test. Multi- CECP with early cancer (T1N0, n = 18) versus 28.4% for
variate proportional hazard Cox regression, structural equa- other CECP (n = 157) after esophagogastrectomies (P =
tion modeling (SEPATH), Monte Carlo simulation, bootstrap 0.000 by log-rank test) (Fig. 5). Early CEC was defined,
simulation and neural networks computing were used to de- based on the final histopathologic report of the resection
termine any significant dependence [15-21]. Neural networks specimen, as tumor limited to the mucosa or submucosa and
computing, system, biometric and statistical analyses were not extending into the muscular wall of the cardioesophageal
conducted using CLASS-MASTER program (Stat Dialog, junction, up to 2 cm in diameter with N0 [12]. All patients
Inc., Moscow, Russia), SANI program (Stat Dialog, Inc., with stage T1N0 did not receive adjuvant chemoimmuno-
Moscow, Russia), DEDUCTOR program (BaseGroup Labs, therapy. Correspondingly, the overall 10-year survival for
Inc., Riazan, Russia), SPSS (SPSS Inc., Chicago, IL, USA), CECP with the early cancer was 92.6% and was significantly
STATISTICA and STATISTICA Neural Networks program better compared to 18.4% for others patients (P = 0.000).
(Stat Soft, Inc., Tulsa, OK, USA), MATHCAD (MathSoft,

5. Cardioesophageal Cancer: Best Treatment Strategies The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 25
Surv iv al F unction
5-Y ear Surv iv al of C ardioes ophageal C anc er Patients =36.1% ; 10-Y ear Surv iv al=26.6%;
n=175
C om plete C ens ored
1.2
1.1
Cumulative Proportion Surviving
1.0
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
-5 0 5 10 15 20 25
Y ears af ter Es ophagogas trec tom y
v
Fig. (4). General cumulative survival of CECP with stage T1-4N0-2M0, n = 175 after radical esophagogastrectomies: cumulative 5-year
survival = 36.1%, 10-year survival = 26.6%.
C um ulativ e Proportion Surv iv ing (Kaplan-Meier)
5-Y ear Surv iv al of E arly C ardioes ophageal C anc er Patients =100% ;
5-Y ear Surv iv al of Inv as iv e C ardioes ophageal C ancer Patients =28.4% ;
P=0.000 by Log-Rank Tes t
C om plete C ensored
1.0
Cumulative Proportion Surviving
0.9
0.8
0.7
0.6
0.5
Inv as iv e C ardioes ophageal C ancer P atients , n=157
0.4 Early C ardioes ophageal C anc er Patients , n=18
0.3
0.2
0.1
0.0
0 5 10 15 20 25
Y ears af ter Es ophagogas trec tom y
u
Fig. (5). Survival of CECP with early cancer (n = 18) was significantly better compared with invasive cancer (n = 157) (P = 0.000 by log-
rank test).

6. 26 The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 Oleg Kshivets
C um ulativ e Proportion Surv iv ing (Kaplan-Meier)
5-Y ear Surv iv al of C ardioesophageal C anc er Patients with N 0=57.1% ;
5-Y ear Surv iv al of C ardioes ophageal C anc er Patients with N 1-2=20.9% ;
P=0.000 by Log Rank Tes t
C om plete C ensored
1.0
0.9 C ardioes ophageal C anc er Patients wit h N 0, n=70
C ardioes ophageal C anc er Patients wit h N 1-2, n=105
0.8
Cumulative Proportion Surviving
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
-0.1
0 5 10 15 20 25
Y ears af t er Es ophagogas trect om y
u
Fig. (6). Survival of CECP with N0 (n = 70) was significantly better compared with N1-2 metastases (n = 105) (P = 0.000 by log-rank test).
Second, we observed relative good 5YS of CECP with blood cell subpopulations and cancer cell population) (Table
N0 (57.1%, n = 70) as compared with 5YS of CECP with 1).
N1-N2 (20.9%, n = 107) after radical procedures (P = 0.000
For comparative purposes, clinicomorphological vari-
by log-rank test) (Fig. 6). Accordingly, the overall 10-year
ables of CECP (n = 157: 53 5-year survivors and 104 losses)
survival for CECP with N0 reached 53% and was signifi-
were tested by neural networks computing (4-layer percep-
cantly superior compared to 10.4% for CECP with lymph tron) (Fig. 8). To obtain a more exact analysis 18 patients
node metastasis. Owing to the relatively high frequency of
being alive less than 5 years after radical procedures without
distant failure after surgical resection of CEC with lymph
relapse were excluded from the sample. Multilayer percep-
nodal metastasis, it has been generally accepted that nodal
tron was trained by Levenberg-Marquardt method. Obvi-
metastasis would be an indicator of systemic metastasis
ously, analyzed data provide significant information about
[12,14]. Consequently, at least two separate subsets of pa-
CEC prediction. High accuracy of classification - 100% (5-
tients can be defined from present study: those with N0 year survivors vs losses) was achieved in analyzed sample
status and those with N1-2 involvement. These factors must
(baseline error = 0.0009e-12, are under ROC curve = 1.0). In
be taken into account in system analysis of CECP survival
other words it remains formally possible that reviled the sev-
and are particularly cogent when attempting to translate ob-
enteen factors might predate neoplastic generalization: phase
tained results into best treatment strategies.
transition early---invasive CEC (rank = 1), phase transition
For 47 CECP in adjuvant chemoimmunotherapy arm N0---N1-2 (rank = 2), adjuvant chemoimmunotherapy (rank
(group A), overall LS was 1844.7 ± 1874.6 days and for 128 = 3), T1-4 (4), gender (5), prothrombin index (6), weight (7),
CECP in the control (group B), overall LS was 1211.5 ± glucose (8), age (9), coagulation time (10), ratio eosino-
1283.3 days (P = 0.000 by log-rank). The overall 5YS of phils/cancer cells (11), erythrocytes/cancer cells (12), hemor-
CECP for group A reached 69.9% and was significantly su- rhage time (13), protein (14), hemoglobin (15), segmented
perior compared to 26.6% for group B (P = 0.000) (Fig. 7). neutrophils/cancer cells (16), stab neutrophils/cancer cells
All parameters were analyzed in a multivariate Cox (17) (Table 2). Genetic algorithm selection and bootstrap
simulation confirmed significant dependence between 5YS
model. In accordance with this Cox model (global 2 =
of CECP after radical procedures and all recognized vari-
126.03; Df = 25; P = 0.000), the nineteen variables signifi-
ables (Tables 3, 4). Moreover, bootstrap simulation con-
cantly explained survival of CECP after surgery: phase tran-
firmed the paramount value of cell ratio factors and the two
sition early---invasive cancer, phase transition N0---N1-N2,
very special patient’s homeostasis states: patients with early
tumor growth, T1-4, adjuvant chemoimmunotherapy, age,
blood cell subpopulations and cell ratio factor (ratio between CEC and N1-2 metastasis.

7. Cardioesophageal Cancer: Best Treatment Strategies The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 27
C um ulativ e Proportion Surv iv ing (Kaplan-Meier)
5-Y ear Surv iv al of C ardioesophageal C anc er Patients af ter A d. C H IT=69.9% ;
5-Y ear Surv iv al of C ardioes ophageal C anc er Patients af ter S urgery alone=26.6% ;
P=0.000 by Log-Rank Tes t
C om plete C ensored
1.0
Cumulative Proportion Surviving
0.9 Only Surgery , n=128
Adjuv ant C hem oim m unotherapy , n=47
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.0
0 5 10 15 20 25
Y ears af ter Es ophagogas trec tom y
u
Fig. (7). Survival of CECP after esophagogastrectomies in group A (adjuvant chemoimmunotherapy) (n = 47) and B (surgery alone) (n =
128). 5-year survival of CECP in group A (69.9%) was significantly better compared with group B (26.6%) (P = 0.000 by log-rank test).
Table 1. Results of Multivariate Proportional Hazard Cox It is necessary to note a very important law: both transi-
Regression Modeling in Prediction of CECP Sur- tions of the early cancer into the invasive cancer, as well as
vival After Esophagogastrectomies (n = 175) the cancer with N0 into the cancer with N1-2, have the phase
character. These results testify by mathematical (Holling-
Variables in the Equation Wald df P Tenner) and imitating modeling of system “CEC-patient
homeostasis” in terms of synergetics (Figs. 9, 10). This also
Phase Transition “Early---Invasive CEC” 3.962 1 0.047 proves the first results received earlier in the work [12].
Phase Transition “N0---N1-2” 11.537 1 0.001 Presence of the two phase transitions is evidently shown on
T 7.957 3 0.047 Kohonen self-organizing neural networks maps (Fig. 11).
T(1) 7.601 1 0.006
T(2) 2.616 1 0.106 All of these differences and discrepancies were further
T(3) 0.604 1 0.437 investigated by structural equation modeling (SEPATH) as
Age 8.358 1 0.004 well as Monte Carlo simulation. From the data, summarized
Adjuvant Chemoimmunotherapy 44.417 1 0.000 in Fig. (12) it could be recognized that the six clusters sig-
Erythrocytes/Cancer Cells 7.321 1 0.007 nificantly predicted 5YS and LS of CECP after complete
Leucocytes/Cancer Cells 6.477 1 0.011 esophagogastrectomies: 1) phase transition “early-invasive
Eosinophils/Cancer Cells 3.604 1 0.058
CEC” (P = 0.000); 2) phase transition “CEC with N0-CEC
Stab Neutrophils/Cancer Cells 4.281 1 0.039
Segmented Neutrophils/Cancer Cells 4.753 1 0.029
with N1-2” (P = 0.001); 3) cell ratio factors (P = 0.000); 4)
Lymphocytes/Cancer Cells 3.108 1 0.078 CEC characteristics (P = 0.000); 5) biochemical homeostasis
Healthy Cells/Cancer Cells 6.273 1 0.012 (P = 0.003); 6) adjuvant chemoimmunotherapy (P = 0.000)
Segmented Neutrophils (%) 8.939 1 0.003 (Fig. 12). It is necessary to pay attention, that both phase
Monocytes (%) 9.349 1 0.002 transitions strictly depend on blood cell circuit (P = 0.000),
Monocytes (abs) 11.445 1 0.001 cell ratio factors (P = 0.000-0.001), biochemical homeostasis
Stab Neutrophils (tot) 7.845 1 0.005 (P = 0.000) and hemostasis system (P = 0.000-0.017).
Leucocytes (tot) 5.683 1 0.017
Eosinophils (tot) 4.038 1 0.044 DISCUSSION
Lymphocytes (tot) 3.264 1 0.071
Monocytes (tot) 3.674 1 0.055
Adequate treatment of CEC is extremely difficult and not
Tumor Growth 17.210 2 0.000 yet problem solving. Firstly, the cardioesophageal cancer
Tumor Growth(1) 14.042 1 0.000 surgery demands virtuosous, very accurate and aggressive
Tumor Growth(2) 5.023 1 0.025 surgical technique, sometime multiorgans resections
Rh-factor 5.960 1 0.015

8. 28 The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 Oleg Kshivets
Fig. (8). Configuration of neural networks: 4-layer perceptron.
Table 2. Results of Neural Networks Computing in Predic- Table 3. Results of Neural Networks Genetic Algorithm Se-
tion of 5-Year Survival of CECP After lection in Prediction of 5-Year Survival of CECP Af-
Esophagogastrectomies (n = 157: 53 5-Year Survi- ter Esophagogastrectomies (n = 157: 53 5-Year Sur-
vors and 104 Losses) vivors and 104 Losses)
Sample n = 157 CECP, n = 157 Useful for 5-Year
NN
NN Factors Rank Factors Survival
Error Ratio
1 Phase Transition “N0---N1-2” Yes
1 Phase Transition “Early---Invasive Cancer” 1 0.539 539e+7 2 Phase Transition “Early---Invasive Cancer” Yes
2 Phase Transition “N0---N1-2” 2 0.426 426e+7
3 Adjuvant Chemoimmunotherapy Yes
3 Adjuvant Chemoimmunotherapy 3 0.329 329e+7
4 Erythrocyte/Cancer Cells Yes
4 T 4 0.300 300e+7
5 Gender 5 0.285 285E+7 5 Leucocytes/Cancer Cells Yes
6 Prothrombin Index 6 0.176 176e+7 6 Eosinophils/Cancer Cells Yes
7 Weight 7 0.155 155e+7 7 Segmented Neutrophils/Cancer Cells Yes
8 Glucose 8 0.122 122e+7 8 Healthy Cells/Cancer Cells Yes
9 Age 9 0.080 799e+6
9 Tumor Size Yes
10 Coagulation Time 10 0.080 798e+6
10 T Yes
11 Eosinophils/Cancer Cells 11 0.080 798e+6
12 Erythrocytes/Cancer Cells 12 0.080 798e+6 11 Hemoglobin Yes
13 Hemorrhage Time 13 0.017 166e+6 12 Coagulation Time Yes
14 Protein 14 0.000 155e+2 13 Hemorrhage Time Yes
15 Hemoglobin 15 0.000 1852.5 14 Prothrombin Index Yes
16 Segmented Neutrophils/Cancer Cells 16 0.000 474.43
15 Glucose Yes
17 Stab Neutrophils/Cancer Cells 17 0.000 1.323
16 Rest Nitrogen Yes
Baseline Error 0.0009e-12
17 Bilirubin Yes
Area under ROC Curve 1.000
18 Gender Yes
Correct Classification Rate (%) 100.00

9. Cardioesophageal Cancer: Best Treatment Strategies The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 29
Table 4. Results of Bootstrap Simulation in Prediction of 5- especially for local advanced CEC for completeness (R0)
Year Survival of CECP After Esophagogastrecto- and always will be the scope of activity of elite top surgeons
mies (n = 157: 53 5-Year Survivors and 104 Losses) [2,3,11,12]. Real surgical removal of tumor, subadjacent
tissues and lymph node metastases remains the cornerstone
Number of of management of this very aggressive cancer giving the real
CECP, n = 157 chance for cure in spite of extensive research over the last 30
Rank Samples = 3333 P<
years in terms of chemotherapy, radiotherapy, immunother-
Significant Factors Kendall’Tau-A
apy and gene therapy [1,3,7]. Nevertheless, the effectiveness
Tumor Size 1 -0.228 0.000 of radical esophagogastrectomies already reached its limit
T 2 -0.223 0.000 and leaves much to be desired: the average real 5-year sur-
Healthy Cells/Cancer Cells 3 0.215 0.000 vival rate of radically operated CECP even after combined
Erythrocytes/Cancer Cells 4 0.211 0.000 multiorgans resections is 20-35% and practically is not im-
Leucocytes/Cancer Cells 5 0.205 0.000 proved during the last 25-30 years, as the great majority of
Lymphocytes/Cancer Cells 6 0.197 0.000 patients has already advanced CEC. Secondly, modern rough
Phase Transition “N0---N1-2” 7 -0.189 0.000 estimated TNM-classification is based only on cancer char-
Segmented Neutrophils/Cancer Cells 8 0.189 0.000 acteristics and does not take into account at all the features of
Thrombocytes/Cancer Cells 9 0.182 0.000 extremely complex alive supersystem - the patient’s organ-
Monocytes/Cancer Cells 10 0.159 0.01 ism. From this it follows that the prediction of CEC is barely
Phase Transition “Early---Invasive Cancer” 11 -0.144 0.01 non adequate. On this point, we used complex system analy-
G 12 -0.137 0.05 sis, artificial intelligence (neural networks computing), simu-
Eosinophils/Cancer Cells 13 0.134 0.05 lation modeling and statistical methods in combination, be-
Adjuvant Chemoimmunotherapy 14 0.122 0.05 cause the different approaches yield complementary pieces
Coagulation Time 15 -0.121 0.05 of prognostic information. Great advantage of the artificial
Histology 16 -0.117 0.05 intelligence methods is the opportunity to find out hidden
Tumor Growth 17 -0.112 0.05
Fig. (9). Results of Holling-Tenner modeling of system “CEC-Lymphocytes” in prediction of CECP survival after esophagogastrectomies
(dynamics of early cancer: Lymphocytes/Cancer Cells = 1/1; dynamics of cancer with N0: Lymphocytes/Cancer Cells = 3/4; dynamics of
cancer with N1-2: Lymphocytes/Cancer Cells = 2/3; cancer generalization: Lymphocytes/Cancer Cells = 1/10).

10. 30 The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 Oleg Kshivets
Fig. (10). Presence of the two phase transitions “early cancer-invasive cancer” and “cancer with N0-cancer with N1-2” in terms of synerget-
ics.
Fig. (11). Results of Kohonen self-organizing neural networks computing in prediction of CECP survival after esophagogastrectomies (n =
157). The black curve line stand for 5-year survivors below and for losses above. The area under more dark-color shadow stand for CECP
with N0 and the area under the weak-colored shadow stand for CECP with N1-2. The area under the less dark-color shadow stand for early
CECP, the area under the weak-colored shadow and more dark-color shadow stand for invasive CECP.

11. Cardioesophageal Cancer: Best Treatment Strategies The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 31
Fig. (12). Significant networks between CECP (n = 157) survival, cancer characteristics, blood cell circuit, cell ratio factors, hemostasis sys-
tem, biochemical homeostasis, anthropometric data, phase transition “early cancer-invasive cancer”, phase transition “cancer with N0-cancer
with N1-2” and adjuvant chemoimmunotherapy (SEPATH network model).
interrelations which cannot be calculated by analytical and cal situation for CECP with N1-2 after radical procedures.
system methods. Meanwhile, huge merit of simulation mod- Present research only confirmed this axiom. In the given
eling is the identification of dynamics of any supersystem, situation high-precision prediction of CECP survival after
including alive supersystem like human homeostasis, on the surgery, which allows to select concrete patients for adjuvant
hole in time [12,15-21]. treatment and to cut huge financial expenses, has a great
value.
For the early CEC only radical surgery is satisfactory
proof: 5YS of patients with early CEC after In conclusion, best treatment strategies for CECP are: 1)
esophagogastrectomies reaches 90-100% without adjuvant screening and early detection of CEC; 2) availability of very
treatment. From this follows the absolute necessity of experienced thoracoabdominal surgeons because of com-
screening and early detection of CEC. plexity of radical procedures; 3) aggressive en block surgery
and adequate abdominal, mediastinal, cervical lymphadenec-
The situation becomes very complicated at once if we
tomy for completeness; 4) high-precision prediction; 5) ad-
have local advanced CEC and, unfortunately, such patients
juvant chemoimmunotherapy for CECP with unfavorable
make up the majority. Without radical procedures these
CECP usually quickly die. Only world top surgeons are ca- prognosis.
pable to perform such combined operations radically. In case REFERENCES
of success 20-30% of patients with locally advanced CEC
[1] Rizzetto C, DeMeester SR, Hagen JA, et al. En bloc esophagec-
live 5 and more years [3,8]. tomy reduces local recurrence and improves survival compared
The most widely accepted treatment strategy for lymph with transhiatal resection after neoadjuvant therapy for esophageal
adenocarcinoma. J Thorac Cardiovasc Surg 2008; 35: 1228-36.
node metastasis is the subsequent initiation of multimodality [2] DeMeester SR. Adenocarcinoma of the esophagus and cardia: a
treatment, including surgery, adjuvant/neoadjuvant chemo- review of the disease and its treatment. Ann Surg Oncol 2006;
therapy or chemoradiation [1,14]. At that radical surgical 13(1): 12-30.
removal of CEC and lymph node metastasis plays a para- [3] Davydov MI, Ter-Ovanesov MD, Stilidi IS, et al. Cancer of the
proximal section of the stomach: the standards of surgical treatment
mount role again, allowing to decrease sharply the number of based on 30 years of experience. Vestn Ross Akad Med Nauk
cancer cell population in patient’ organism and to warn pos- 2002; 1: 25-9.
sible deadly complications (e.g., profuse hemorrhage). Theo- [4] Gee DW, Rattner DW. Management of gastroesophageal tumors.
retically chemoimmunotherapy is the most effective when Oncologist 2007; 12: 175-85.
used in patients with a relatively low residual malignant cell [5] Siewert JR, Holscher AH, Becker K, et al. Kardiakarzinom: Ver-
such einer therapeutisch relevanten Klassifikation. Chirurg 1996;
population (approximately 1 billion cancer cells per patient) 58: 25-34.
in terms of hidden micrometastasis [12]. This is typical clini- [6] Siewert JR, Feith M, Werner M, Stein HJ. Adenocarcinoma of the
esophagogastric junction. Results of surgical therapy based on ana-

12. 32 The Open Cardiovascular and Thoracic Surgery Journal, 2009, Volume 2 Oleg Kshivets
tomical /topographic classification in 1002 consecutive patients. [13] Morozow VG, Chavinson VC. Isolation, refinement and identifica-
Ann Surg 2000; 232: 353-61. tion of immunomodulated polypeptide from calf and human thy-
[7] Barbour AP, Rizk NP, Gonen M, et al. Adenocarcinoma of the mus. Biochemistry 1981; 9: 1652-9.
gastroesophageal junction: influence of esophageal resection mar- [14] Kshivets O. Esophageal cancer: Optimization of management.
gin and operative approach on outcome. Ann Surg 2007; 246: 1-8. Open Cardiovasc Thorac Surg J 2008; 1: 1-11.
[8] Pedrazzani C, de Manzoni G, Marrelli D, et al. Lymph node in- [15] Bazykin AD. Mathematical biophysics of cooperating populations.
volvement in advanced gastroesophageal junction adenocarcinoma. Science: Moscow 1985; p. 181.
J Thorac Cardiovasc Surg 2007; 134: 378-85. [16] Haken H. Information and self-organization. A macroscopic ap-
[9] Shen KR, Cassiva St D, Deschamps C, et al. Surgical treatment of proach to complex systems. Springer: Berlin 2006; p. 240.
tumours of the proximal stomach with involvement of the distal [17] Odom-Maryon T. Biostatistical methods in oncology. Cancer man-
esophagus: A 26-years experience with Siewert type III tumours. J agement: A multidisciplinary approach. 1st ed. Huntington, NY:
Thorac Cardiovasc Surg 2006; 132: 755-62. PRP Inc 1996; 788-802.
[10] Nakamura T, Oguma H, Sasagawa T, et al. Left thoracoabdominal [18] Mirkin B.G. A sequential fitting procedure for linear data analysis
approach for adenocarcinoma of the esophagogastric junction. models. J Classification 1990; 7: 167-96.
Hepatogastroenterology 2008; 55: 1332-7. [19] Joreskog KG, Sorbom D. Recent development in structural equa-
[11] Yonemura Y, Kojima N, Kawamura T, et al. Treatment results of tion modeling. J Mark Res 1982; 19: 404-16.
adenocarcinoma of the gastroesophageal junction. Hepatogastroen- [20] Bostwick DG, Burke HB. Prediction of individual patient outcome
terology 2008; 55: 475-81. in cancer: comparison of artificial neural networks and Kaplan-
[12] Kshivets O. Expert system in diagnosis and prognosis of malignant Meier methods. Cancer 2001; 91(8): 1643-6.
neoplasms. Dissertation for Sc.D. Tomsk 1995; p. 486. [21] Husmeier D. The Bayesian evidence scheme for regularizing prob-
ability-density estimating neural networks. Neural Comput 2000;
12(11): 2685-717.
Received: February 12, 2009 Revised: February 18, 2009 Accepted: February 18, 2009
© Oleg Kshivets; Licensee Bentham Open.
This is an open access article licensed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-
nc/3.0/) which permits unrestricted, non-commercial use, distribution and reproduction in any medium, provided the work is properly cited.